Newsgroups: php.internals
Path: news.php.net
Xref: news.php.net php.internals:111393
MIME-Version: 1.0
References: <alpine.DEB.2.23.453.2008041440380.7489@singlemalt.home.derickrethans.nl>
 <AI3wuqzewCx8eX9KNz6ZwnMkyvcYwttIVkZsWWHHv7U@dedadlershop92.panaxis.li>
In-Reply-To: <AI3wuqzewCx8eX9KNz6ZwnMkyvcYwttIVkZsWWHHv7U@dedadlershop92.panaxis.li>
Date: Sun, 9 Aug 2020 17:37:44 +0200
Message-ID: <CAKy=g0ojQ1bQqiiUYZfDXOZZi5Wpii8UOYuXZOosBKk2igy_WQ@mail.gmail.com>
To: PHP Developers Mailing List <internals@lists.php.net>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Subject: Re: [PHP-DEV] [RFC] Shorter Attribute Syntax Change RFC 0.2
From: k142es@gmail.com (Kamil ES)

Hi everyone. I=E2=80=99m not a PHP internal, just a modest PHP developer. B=
ut I felt
a desire to share my observation on =E2=80=9C@@=E2=80=9D.

Some symbols looks very okay when doubled. For example, we use =E2=80=9C//=
=E2=80=9D for comments
and =E2=80=9C||=E2=80=9D for logical alternative. They are okay, because th=
ey contains only
two parallel lines. There is a symbol like that in mathematics: =E2=80=9C=
=E2=88=A5=E2=80=9D. Programmer
fonts (for example Fira Code) convert doubled characters to an appropriate
ligature.

On the other side, the at sign =E2=80=9C@=E2=80=9D contains the small lette=
r =E2=80=9Ca=E2=80=9D in a circle.
It=E2=80=99s feels full to the brim. It=E2=80=99s complete and closed whole=
. It=E2=80=99s like a donut
with tasty filling. ;-) But when we double it, the resulting =E2=80=9C@@=E2=
=80=9D is bloated.
Even intuitively not right.

Why =E2=80=9C@@=E2=80=9D was proposed? Because =E2=80=9C@=E2=80=9D is alrea=
dy taken. Sorry, but it=E2=80=99s born out of
helplessness (in the most visible way in comparison to other syntaxes). And
maybe this is the reason why some of community members feel bad about it
(no matter of technical arguments).

One could argue that it=E2=80=99s a matter of taste. Well, only to some deg=
ree. Mankind
invented typography and editing, so rules of esthetic and harmonious writin=
g
exist. PHP has PSR-12 (=E2=80=9CExtended Coding Style=E2=80=9D) for this.

Now I=E2=80=99d like to draw your attention to this syntax:

<<
  @Assert\NotNull
  @Assert\Type('string')
  @ORM\Column(type=3D'string')
>>
private string $username;

It has some good points. First of all, it uses single =E2=80=9C@=E2=80=9D. =
This is a popular
choice in other programming languages, including JavaScript =E2=80=93 a nat=
ural
companion of PHP in Web programming.
\
Nesting attributes in this syntax is very natural.

The mental model of the syntax reads like that: Here I put =E2=80=9C<<=E2=
=80=9D and it starts
a metadata (metaprogramming) context. Now I put =E2=80=9C@=E2=80=9D for eve=
ry attribute. Let=E2=80=99s
finish it with =E2=80=9C>>=E2=80=9D. I like this separation of metaprogrami=
ng and a real code.
It=E2=80=99s something that remainds me DocBlock.

The syntax has it=E2=80=99s weak points, of course. For example, a single a=
ttribute
looks a bit verbose. A good syntax highliting may help to blunt this impres=
sion.

<<@Deprecated>>
public function doSomething()

Moreover, the more spaced syntax may be preffered (and I like it more):

<<
  @Deprecated
>>
public function doSomething()

Another option would be to use a keyword, just like =E2=80=9Cnew=E2=80=9D w=
ith a class name.

meta Assert\NotNull
meta Assert\Type('string')
meta ORM\Column(type=3D'string')
private string $username;

Going back to nested attributes, what I don=E2=80=99t like about =E2=80=9C@=
[]=E2=80=9D and =E2=80=9C#[]=E2=80=9D is that
the at sing is at first before brackets, but for nested attributes it will =
be
put together with an attribute name:

@[Attribute(@NestedAttribute('text')]

I believe this RFC should have an example section to show every syntax in
practice.

Best wishes to all of you.